survSplit: further exploration and related topics

To Danardonos concern of splitting time for records with delayed entry:

There was actually no intention of supporting counting-process input, but 
it turns out to be a one-line change: from
        starttime <- c(starttime, rep(cut, each = n))
to
        starttime<-c(starttime, pmax(starttime, rep(cut,each=n)))

about 20 lines down from the top.  This will be in the next release of 
survival, but not until after R 2.0.1


 	-thomas